The purpose of this project was to investigate nonhuman primates' understanding of the nature of joystick-mediated testing procedures. These tasks require subjects to manipulate a joystick and cause cursor images to contact of stationary and/or moving targets. While many species of nonhuman primates have been trained to reliably perform these tasks, little research has focused on what the subjects actually understand when they perform these tasks. In our studies, subjects were trained to manipulate a joystick and cause a self-controlled image (SCI) to contact a moving target on a computer screen. This task was made competitive by adding a third, computer-controlled image (CCI) which also moved toward the target. Task difficulty was increased by making the CCI movements more efficient and by making the CCI and SCI images visually identical. Both species were capable of moving the SCI into the target before the CCI on a significant majority of trials. However, under the most difficult conditions, only the rhesus monkeys continued to perform above chance. Successful completion of this task suggests subjects have an understanding of which computer image they control during the task. A comparison with different simulated response strategies indicated that each species performed better than random movement simulations but only rhesus monkeys were comparable to simulations that could discriminate between the CCI and SCI. Further research will attempt to clarify the possible confounds of training history by testing relatively nave subjects on the same task.